Patchwork macvlan: receive multicast with local address

login
register
mail settings
Submitter stephen hemminger
Date Nov. 2, 2011, 10:11 p.m.
Message ID <20111102151153.028e222f@nehalam.linuxnetplumber.net>
Download mbox | patch
Permalink /patch/123353/
State Accepted
Delegated to: David Miller
Headers show

Comments

stephen hemminger - Nov. 2, 2011, 10:11 p.m.
When implementing VRRP v2 using macvlan several problems were
discovered.  VRRP is weird in that all routers participating
in a redundant group use the same virtual MAC address.
Macvlan is a natural driver to use for this but it doesn't
work.  The problem is that packets with a macvlan device's
source address are not received.

The problem is actually a regression that date back almost 2 years now.
The original problems started with:

commit 618e1b7482f7a8a4c6c6e8ccbe140e4c331df4e9
Author: Arnd Bergmann <arnd@arndb.de>
Date:   Thu Nov 26 06:07:10 2009 +0000

    macvlan: implement bridge, VEPA and private mode
    
This patches restores the original 2.6.32 behavior. Allowing multicast
packets received with the VRRP source address to be received.


Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>

---
This should go to -net, and -stable (since it is a regression).

P.s: The VEPA patch also introduced another regression, it changed
the default mode from private to VEPA which also breaks application
assumptions. But changing it back after 2 years also risks also breaking
things.

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller - Nov. 4, 2011, 9:40 p.m.
From: Stephen Hemminger <shemminger@vyatta.com>
Date: Wed, 2 Nov 2011 15:11:53 -0700

> When implementing VRRP v2 using macvlan several problems were
> discovered.  VRRP is weird in that all routers participating
> in a redundant group use the same virtual MAC address.
> Macvlan is a natural driver to use for this but it doesn't
> work.  The problem is that packets with a macvlan device's
> source address are not received.
> 
> The problem is actually a regression that date back almost 2 years now.
> The original problems started with:
> 
> commit 618e1b7482f7a8a4c6c6e8ccbe140e4c331df4e9
> Author: Arnd Bergmann <arnd@arndb.de>
> Date:   Thu Nov 26 06:07:10 2009 +0000
> 
>     macvlan: implement bridge, VEPA and private mode
>     
> This patches restores the original 2.6.32 behavior. Allowing multicast
> packets received with the VRRP source address to be received.
> 
> Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>

Applied and queued up for -stable, thanks.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

--- a/drivers/net/macvlan.c	2011-11-02 13:12:52.186069720 -0700
+++ b/drivers/net/macvlan.c	2011-11-02 14:50:54.476670346 -0700
@@ -192,6 +192,13 @@  static rx_handler_result_t macvlan_handl
 			 */
 			macvlan_broadcast(skb, port, src->dev,
 					  MACVLAN_MODE_VEPA);
+		else {
+			/* forward to original port. */
+			vlan = src;
+			ret = macvlan_broadcast_one(skb, vlan, eth, 0);
+			goto out;
+		}
+
 		return RX_HANDLER_PASS;
 	}